A / B testing method for behavior characteristics of autonomous vehicle and related device

Abstract

The application discloses an A / B test method for behavior characteristics of an automatic driving vehicle and related equipment, and the method comprises the following steps: acquiring first vehicle road test data added with behavior characteristics and second vehicle road test data without the behavior characteristics; performing scene simulation on the first vehicle road test data and the second vehicle road test data to obtain first vehicle scene video and second vehicle scene video; fusing the second vehicle scene video based on the first vehicle scene video to obtain fused scene video; and displaying the behavior difference between the two main vehicles in the two scene videos. It can be seen that, since the frame numbers of the two vehicle scene videos are the same, the two vehicle scene videos can be fused according to each frame, so that the developer can observe the two main vehicles in the same video, without observing the main vehicles of the two vehicle scene videos twice separately, thereby guaranteeing the efficiency and accuracy of judging the difference between the simulation scene results before and after the automatic driving vehicle is endowed with the behavior characteristics.

Description

Technical Field

[0001] This application relates to the field of testing technology, and more specifically, to an A / B testing method and related equipment for the behavioral characteristics of autonomous vehicles. Background Technology

[0002] With the development of intelligent technologies, autonomous driving has been extensively researched and gradually implemented in real life. Due to the extremely high safety requirements of autonomous driving, multiple tests are necessary during the development phase of autonomous vehicles. Typically, after assigning a behavioral characteristic to an autonomous vehicle, developers need to conduct A / B testing on that behavioral characteristic. A / B testing involves simulating road test data of an autonomous vehicle with the assigned behavioral characteristic and road test data of an autonomous vehicle without that characteristic, respectively, to obtain simulation scenario results. Developers can then observe the differences in the simulation scenario results before and after the autonomous vehicle is assigned the behavioral characteristic.

[0003] However, currently, developers rely solely on human eyes to judge the differences in simulation results before and after autonomous vehicles are given behavioral characteristics. Since the performance of road test behavior is greatly affected by environmental factors and is abstract, the efficiency of developers observing with human eyes is low, and long-term observation of the differences can easily lead to eye fatigue, resulting in low accuracy in judging the differences in simulation results before and after autonomous vehicles are given behavioral characteristics. Summary of the Invention

[0004] In view of the above problems, this application is proposed to provide an A / B testing method and related equipment for the behavioral characteristics of autonomous vehicles, so as to improve the accuracy of judging the difference between the simulation scene results before and after the autonomous vehicle is given behavioral characteristics.

[0005] To achieve the above objectives, the following specific solutions are proposed:

[0006] An A / B testing method for the behavioral characteristics of autonomous vehicles includes:

[0007] Acquire first vehicle road test data of an autonomous vehicle with added behavioral features, and second vehicle road test data of an autonomous vehicle without added behavioral features;

[0008] Scene simulations were performed on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video. The number of frames in the second vehicle scene video was the same as the number of frames in the first vehicle scene video.

[0009] Using the first vehicle scene video as a reference video, the second vehicle scene video is fused to obtain a fused scene video, which displays the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video.

[0010] Optionally, the step of performing scene simulations on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video includes:

[0011] The road test data of the first vehicle is input into the scene simulation system, and the scene video of the first vehicle is output.

[0012] The second vehicle road test data is input into the scene simulation system, and the second vehicle scene video is output.

[0013] Optionally, using the first vehicle scene video as a reference video, the second vehicle scene video is fused to obtain a fused scene video, including:

[0014] The main vehicle and obstacles in the first vehicle scene video are marked first;

[0015] The main vehicle and obstacles in the second vehicle scene video are marked with a second mark, and the mark color of the second mark is different from the mark color of the first mark;

[0016] For each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, all traffic elements, road elements and obstacles in the first target frame are retained, and traffic elements, road elements and obstacles in the second target frame that are different from those in the first target frame are superimposed on the first target frame to obtain the first fused frame of the first target frame.

[0017] For each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, when the coordinate position of the main vehicle in the second target frame is different from the coordinate position of the main vehicle in the first target frame, the main vehicle in the second target frame is superimposed on the first target frame to obtain the second target frame.

[0018] The second fusion frames of each first target frame are combined to obtain a fused scene video.

[0019] Optionally, the method further includes:

[0020] Acquire the indicator light switch status information, gear shift information, and acceleration / deceleration information of the main vehicle in each first target frame, and the indicator light switch status information, gear shift information, and acceleration / deceleration information of the main vehicle in each second target frame;

[0021] For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the indicator light status information of the main vehicle in the first target frame is different from the indicator light status information of the main vehicle in the second target frame, a third mark is marked on both the main vehicle in the first target frame and the main vehicle in the second target frame. The mark color of the third mark is different from the mark color of the first mark and the mark color of the second mark.

[0022] Optionally, the method further includes:

[0023] For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the gear shifting information of the main vehicle in the first target frame is different from the gear shifting information of the main vehicle in the second target frame, a fourth mark is marked on both the main vehicle in the first target frame and the main vehicle in the second target frame. The mark color of the fourth mark is different from the mark color of the first mark, the mark color of the second mark and the mark color of the third mark.

[0024] Optionally, the method further includes:

[0025] For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the acceleration / deceleration information of the main vehicle in the first target frame is different from that of the main vehicle in the second target frame, a fifth mark is marked on both the main vehicle in the first target frame and the main vehicle in the second target frame. The mark color of the fifth mark is different from the mark color of the first mark, the mark color of the second mark, the mark color of the third mark, and the mark color of the fourth mark.

[0026] Optionally, the main vehicle and obstacles in the first vehicle scene video are marked first, including:

[0027] In each frame of the first vehicle scene video, when the distance between the main vehicle and the target obstacle in that frame is less than a preset spacing, the main vehicle and the target obstacle in that frame are marked for the first time.

[0028] Optionally, a second label is applied to the main vehicle and obstacles in the second vehicle scene video, including:

[0029] In each frame of the second vehicle scene video, when the distance between the main vehicle and the target obstacle in that frame is less than a preset spacing, the main vehicle and the target obstacle in that frame are marked a second time.

[0030] Optionally, the method further includes:

[0031] Obtain the speed information, acceleration information, and steering wheel information of the main vehicle in each frame of the first vehicle scene video;

[0032] Obtain the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the second vehicle scene video;

[0033] A behavior feature table is constructed based on the speed, acceleration, and steering information of the main vehicle in each frame of the first vehicle scene video, and the speed, acceleration, and steering information of the main vehicle in each frame of the second vehicle scene video.

[0034] An A / B testing device for the behavioral characteristics of an autonomous vehicle, comprising:

[0035] The road test data acquisition unit is used to acquire first vehicle road test data of an autonomous vehicle with added behavioral features, and second vehicle road test data of an autonomous vehicle without added behavioral features.

[0036] The scene video acquisition unit is used to perform scene simulation on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video, wherein the number of frames of the second vehicle scene video is the same as the number of frames of the first vehicle scene video.

[0037] The scene video fusion unit is used to fuse the second vehicle scene video with the first vehicle scene video as the reference video to obtain a fused scene video, so as to display the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video.

[0038] Optionally, the scene video acquisition unit includes:

[0039] The first vehicle scene video acquisition unit is used to input the first vehicle road test data into the scene simulation system and output the first vehicle scene video.

[0040] The second vehicle scene video acquisition unit is used to input the second vehicle road test data into the scene simulation system and output the second vehicle scene video.

[0041] Optionally, the scene video fusion unit includes:

[0042] The first marking unit is used to mark the main vehicle and obstacles in the first vehicle scene video;

[0043] The second marking unit is used to mark the main vehicle and obstacles in the second vehicle scene video, and the marking color of the second marking is different from the marking color of the first marking.

[0044] The first fusion frame determination unit is used to retain all traffic elements, road elements and obstacles in the first target frame for each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, and to superimpose the traffic elements, road elements and obstacles in the second target frame that are different from those in the first target frame onto the first target frame to obtain the first fusion frame of the first target frame.

[0045] The second fusion frame determination unit is used to, for each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, when the coordinate position of the main vehicle in the second target frame is different from the coordinate position of the main vehicle in the first target frame, superimpose the main vehicle in the second target frame onto the first fusion frame to obtain the second fusion frame of the first target frame.

[0046] The video synthesis unit is used to synthesize the second fusion frames of each first target frame to obtain a fused scene video.

[0047] Optionally, the device may also include:

[0048] The main vehicle information acquisition unit is used to acquire the indicator light switch status information, gear shift information and acceleration / deceleration information of the main vehicle in each first target frame, and the indicator light switch status information, gear shift information and acceleration / deceleration information of the main vehicle in each second target frame.

[0049] The third marking unit is used to mark the main vehicle in the first target frame and the main vehicle in the second target frame in the second vehicle scene video with a third mark if the indicator light status information of the main vehicle in the first target frame is different from that of the main vehicle in the second target frame. The marking color of the third mark is different from that of the marking color of the first mark and the marking color of the second mark.

[0050] Optionally, the device may also include:

[0051] The fourth marking unit is used to mark the main vehicle in the first target frame and the main vehicle in the second target frame in the second vehicle scene video with a fourth mark if the gear shifting information of the main vehicle in the first target frame is different from that of the main vehicle in the second target frame. The marking color of the fourth mark is different from the marking color of the first mark, the marking color of the second mark and the marking color of the third mark.

[0052] Optionally, the device may also include:

[0053] The fifth marking unit is used to mark the main vehicle in the first target frame and the main vehicle in the second target frame in the second vehicle scene video with a fifth mark if the acceleration and deceleration information of the main vehicle in the first target frame is different from that in the second target frame. The marking color of the fifth mark is different from the marking color of the first mark, the marking color of the second mark, the marking color of the third mark and the marking color of the fourth mark.

[0054] Optionally, the first marking unit includes:

[0055] The first vehicle obstacle marking unit is used to mark the main vehicle and the target obstacle in each frame of the first vehicle scene video when the distance between the main vehicle and the target obstacle in the frame is less than a preset distance.

[0056] Optionally, the second marking unit includes:

[0057] The second vehicle obstacle marking unit is used to mark the main vehicle and the target obstacle in each frame of the second vehicle scene video when the distance between the main vehicle and the target obstacle in the frame is less than a preset distance.

[0058] Optionally, the device may also include:

[0059] The first main vehicle information acquisition unit is used to acquire the speed information, acceleration information and steering wheel information of the main vehicle in each frame of the first vehicle scene video.

[0060] The second main vehicle information acquisition unit is used to acquire the speed information, acceleration information and steering wheel information of the main vehicle in each frame of the second vehicle scene video.

[0061] The behavior feature table construction unit is used to construct a behavior feature table based on the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the first vehicle scene video, and the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the second vehicle scene video.

[0062] An A / B testing device for the behavioral characteristics of an autonomous vehicle, comprising a memory and a processor;

[0063] The memory is used to store programs;

[0064] The processor is used to execute the program to implement the various steps of the A / B testing method for the behavioral characteristics of autonomous vehicles as described above.

[0065] A storage medium storing a computer program, which, when executed by a processor, implements the steps of the A / B testing method for the behavioral characteristics of an autonomous vehicle as described above.

[0066] By employing the above technical solution, this application acquires first vehicle road test data of an autonomous vehicle with added behavioral features and second vehicle road test data of an autonomous vehicle without added behavioral features. Scene simulation is then performed on the first and second vehicle road test data respectively to obtain a first vehicle scene video and a second vehicle scene video. The second vehicle scene video has the same number of frames as the first vehicle scene video. Furthermore, using the first vehicle scene video as a reference video, the second vehicle scene video is fused to obtain a fused scene video, displaying the behavioral differences between the main vehicle in the first and second vehicle scene videos. Therefore, since scene simulation is performed on both vehicle road test data with and without added behavioral features, the resulting two vehicle scene videos have the same number of frames. This allows for the fusion of the two vehicle scene videos frame-by-frame, enabling developers to observe the main vehicle in both the first and second vehicle scene videos within a single video, eliminating the need to observe the main vehicle in each video separately. This facilitates efficient observation of the differences between the two main vehicles and ensures the accuracy of judging the differences in simulation scene results before and after the autonomous vehicle is given behavioral features. Attached Figure Description

[0067] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0068] Figure 1 A schematic diagram of a process for implementing A / B testing of the behavioral characteristics of autonomous vehicles provided in an embodiment of this application;

[0069] Figure 2 A schematic diagram of a device structure for implementing A / B testing of the behavioral characteristics of autonomous vehicles according to an embodiment of this application;

[0070] Figure 3 This is a schematic diagram of the structure of a device for A / B testing of the behavioral characteristics of autonomous vehicles, provided in an embodiment of this application. Detailed Implementation

[0071] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0072] The proposed solution can be implemented based on a terminal with data processing capabilities, such as a computer, server, or cloud application.

[0073] Next, combined Figure 1 The A / B testing method for the behavioral characteristics of autonomous vehicles described in this application may include the following steps:

[0074] Step S110: Obtain first vehicle road test data of autonomous vehicles with added behavioral features, and second vehicle road test data of autonomous vehicles without added behavioral features.

[0075] Specifically, the behavioral characteristics can represent a conscious behavior that an autonomous vehicle exhibits in road traffic, such as avoiding obstacles in advance. The first vehicle road test data can represent vehicle data recorded by the autonomous vehicle while driving in road traffic after the behavioral characteristics have been incorporated, while the second vehicle road test data can represent vehicle data recorded by the autonomous vehicle while driving in road traffic without the incorporated behavioral characteristics.

[0076] Step S120: Perform scene simulation on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video.

[0077] Understandably, because developers need to observe the differences between autonomous vehicles before and after they are given behavioral characteristics, vehicle road test data is not very intuitive. It is necessary to apply vehicle road test data to specific traffic scenarios to obtain more intuitive scene videos.

[0078] The second vehicle scene video has the same number of frames as the first vehicle scene video.

[0079] Specifically, the data format of both the first vehicle scene video and the second vehicle scene video can be MP4 format.

[0080] Step S130: Using the first vehicle scene video as the reference video, fuse the second vehicle scene video to obtain a fused scene video, so as to display the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video.

[0081] Specifically, after merging the second vehicle scene video into the first vehicle scene video, the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video will both be displayed in the same scene. If the two main vehicles do not overlap in a certain frame of the merged scene video, it can be said that the two main vehicles have behavioral differences. Conversely, if only one main vehicle is displayed in a certain frame of the merged scene video (because the positions of the two main vehicles will overlap), it can be said that the two main vehicles do not have behavioral differences.

[0082] The A / B testing method for the behavioral characteristics of autonomous vehicles provided in this embodiment acquires first vehicle road test data of an autonomous vehicle with added behavioral characteristics and second vehicle road test data of an autonomous vehicle without added behavioral characteristics. Scene simulations are then performed on the first and second vehicle road test data to obtain a first vehicle scene video and a second vehicle scene video. The second vehicle scene video has the same number of frames as the first vehicle scene video. Further, using the first vehicle scene video as a reference video, the second vehicle scene video is fused to obtain a fused scene video, which displays the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video. Therefore, since the two vehicle scene videos obtained by simulating the road test data with and without behavioral features are the same number of frames, the two vehicle scene videos can be fused according to each frame. This allows developers to observe the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video in one video, without having to observe the main vehicle in the two vehicle scene videos separately. This helps to efficiently observe the differences between the two main vehicles and ensures the accuracy of judging the difference between the simulation scene results before and after the autonomous vehicle is given behavioral features.

[0083] In some embodiments of this application, the process of performing scene simulations on the first vehicle road test data and the second vehicle road test data to obtain the first vehicle scene video and the second vehicle scene video is described. This process may include:

[0084] S1. Input the road test data of the first vehicle into the scene simulation system and output the scene video of the first vehicle.

[0085] Specifically, the scenario simulation system can accept vehicle road test data as input and simulate the process of a vehicle driving on a traffic road based on the vehicle road test data.

[0086] Before inputting the first vehicle road test data into the scene simulation system, the road scene to be simulated by the scene simulation system can be pre-configured.

[0087] S2. Input the second vehicle road test data into the scene simulation system and output the second vehicle scene video.

[0088] Specifically, the scenario simulation system performs scenario simulation on the second vehicle's road test data, which can be based on the same simulated road scenario as the first vehicle's road test data.

[0089] In some embodiments of this application, the process of fusing the second vehicle scene video with the first vehicle scene video as the reference video to obtain a fused scene video, as mentioned in the above embodiments, is described. This process may include:

[0090] S1. Mark the main vehicle and obstacles in the first vehicle scene video.

[0091] Specifically, in each frame of the first vehicle scene video, when the distance between the main vehicle and the target obstacle in that frame is less than a preset distance, a first mark is made on the main vehicle and the target obstacle in that frame to indicate that the main vehicle and the target obstacle may be at risk of collision.

[0092] The target obstacle can be defined as an obstacle that is less than a preset distance from the main vehicle, and the preset distance can be defined as the critical distance between the main vehicle and the obstacle.

[0093] S2. Mark the main vehicle and obstacles in the second vehicle scene video.

[0094] Specifically, in each frame of the second vehicle scene video, when the distance between the main vehicle and the target obstacle in that frame is less than a preset distance, a second mark is made on the main vehicle and the target obstacle in that frame to indicate that the main vehicle and the target obstacle may be at risk of collision.

[0095] The target obstacle can be defined as an obstacle that is less than a preset distance from the main vehicle, and the preset distance can be defined as the critical distance between the main vehicle and the obstacle. The marker color of the second marker can be different from that of the first marker to distinguish the main vehicle and obstacles in the first vehicle scene video.

[0096] S3. For each first target frame of the first vehicle scene video and the second target frame corresponding to the first target frame in the second vehicle scene video, retain all traffic elements, road elements and obstacles in the first target frame, and superimpose the traffic elements, road elements and obstacles in the second target frame that are different from those in the first target frame onto the first target frame to obtain the first fused frame of the first target frame.

[0097] Specifically, traffic elements can include traffic lights, traffic signs, etc., road elements can include road lines, road outlines, etc., and obstacles can include roadbeds, moving vehicles, stationary vehicles, etc.

[0098] Understandably, since the number of frames in the first vehicle scene video is the same as that in the second vehicle scene video, the frames in both videos can be pre-numbered according to frame order. For the first and second target frames with the same number, the first target frame can be fixed, and background elements (such as traffic elements, road elements, and obstacles) from the second target frame that differ from the first target frame can be superimposed onto the first target frame, thereby achieving the fusion of the two images. The fused image frame retains the features of both frames.

[0099] S4. For each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, when the coordinate position of the main vehicle in the second target frame is different from the coordinate position of the main vehicle in the first target frame, the main vehicle in the second target frame is superimposed on the first target frame to obtain the second target frame.

[0100] It is understandable that when the position of the main vehicle in the first target frame is different from that in the second target frame, it can be said that there is a difference in driving behavior between the main vehicle after adding behavioral features and the main vehicle without adding behavioral features. Therefore, the main vehicle in the second target frame can be superimposed on the first fused frame to obtain the second fused frame of the first target frame. Then, the main vehicle in the first target frame and the main vehicle in the second target frame can coexist in the second fused frame.

[0101] S5. Combine the second fusion frames of each first target frame to obtain a fused scene video.

[0102] The A / B testing method for the behavioral characteristics of autonomous vehicles provided in this embodiment superimposes traffic elements, road elements, obstacles, and the main vehicle of the second target frame onto the first target frame, enabling developers to compare the behavioral differences between the two main vehicles in one frame, thereby improving the accuracy of judging the differences in simulation scene results before and after the autonomous vehicle is given behavioral characteristics.

[0103] In some embodiments of this application, considering that the vehicle road test data stores the indicator light switch data of the main vehicle, when performing scene simulation on the vehicle road test data, the changes in the indicator light switch of the main vehicle can be reflected in the vehicle scene video. Based on this, the A / B testing method for the behavioral characteristics of autonomous vehicles provided in this application may further include:

[0104] S1. Obtain the indicator light switch status information, gear shift information, and acceleration / deceleration information of the main vehicle in each first target frame, and the indicator light switch status information, gear shift information, and acceleration / deceleration information of the main vehicle in each second target frame.

[0105] Specifically, the indicator light switch status information, gear shift information, and acceleration / deceleration information of the main vehicle in each first target frame can be extracted from the first vehicle road test data, and the indicator light switch status information, gear shift information, and acceleration / deceleration information of the main vehicle in each second target frame can be extracted from the second vehicle road test data.

[0106] S2. For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the indicator light status information of the main vehicle in the first target frame is different from the indicator light status information of the main vehicle in the second target frame, a third mark is made on both the main vehicle in the first target frame and the main vehicle in the second target frame.

[0107] It is understood that the marker color of the first marker is used to highlight the main vehicle and obstacles in the first vehicle scene video, the marker color of the second marker is used to highlight the main vehicle and obstacles in the second vehicle scene video, and the marker color of the third marker is used to highlight the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video. Due to the different indicator light states in the same frame, the marker color of the third marker is different from the marker colors of the first marker and the second marker.

[0108] In some embodiments of this application, considering that the vehicle road test data stores the gear shifting data of the main vehicle, when performing scene simulation on the vehicle road test data, the gear shifting changes of the main vehicle can be reflected in the vehicle scene video. Based on this, the A / B testing method for the behavioral characteristics of autonomous vehicles provided in this application may further include:

[0109] For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the gear shifting information of the main vehicle in the first target frame is different from the gear shifting information of the main vehicle in the second target frame, a fourth mark is made on both the main vehicle in the first target frame and the main vehicle in the second target frame.

[0110] It is understood that the color of the first marker is used to highlight the main vehicle and obstacles in the first vehicle scene video, the color of the second marker is used to highlight the main vehicle and obstacles in the second vehicle scene video, the color of the third marker is used to highlight the difference in indicator light status between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video in the same frame, and the color of the fourth marker is used to highlight the difference in gear engagement status between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video in the same frame. Therefore, the color of the fourth marker is different from the colors of the first, second, and third markers.

[0111] In some embodiments of this application, considering that the vehicle road test data stores the acceleration and deceleration data of the main vehicle, when performing scene simulation on the vehicle road test data, the acceleration and deceleration changes of the main vehicle can be reflected in the vehicle scene video. Based on this, the A / B testing method for the behavioral characteristics of autonomous vehicles provided in this application may further include:

[0112] For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the acceleration / deceleration information of the main vehicle in the first target frame is different from that of the main vehicle in the second target frame, the main vehicle in the first target frame and the main vehicle in the second target frame are both marked with a fifth mark.

[0113] It is understood that the color of the first marker is used to highlight the main vehicle and obstacles in the first vehicle scene video; the color of the second marker is used to highlight the main vehicle and obstacles in the second vehicle scene video; the color of the third marker is used to highlight the difference in indicator light status between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video within the same frame; the color of the fourth marker is used to highlight the difference in gear engagement status between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video within the same frame; and the color of the fifth marker is used to highlight the difference in acceleration and deceleration status between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video within the same frame. Therefore, the color of the fifth marker is different from the colors of the first, second, third, and fourth markers.

[0114] In some embodiments of this application, considering that the vehicle road test data stores the speed data, acceleration data, and steering wheel data of the main vehicle, these data are important parameters before and after the autonomous vehicle is given behavioral characteristics. These important parameters can be plotted in a table to allow developers to more intuitively judge the differences between two main vehicles. Based on this, the A / B testing method for the behavioral characteristics of autonomous vehicles provided in this application may further include:

[0115] S1. Obtain the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the first vehicle scene video.

[0116] Specifically, the speed, acceleration, and steering information of the main vehicle in each frame of the first vehicle scene video can be extracted from the first vehicle road test data.

[0117] S2. Obtain the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the second vehicle scene video.

[0118] Specifically, the speed, acceleration, and steering information of the main vehicle in each frame of the second vehicle scene video can be extracted from the second vehicle road test data.

[0119] S3. Construct a behavior feature table based on the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the first vehicle scene video, and the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the second vehicle scene video.

[0120] Specifically, the behavior feature table records the speed, acceleration, and steering information of the two main vehicles in each frame in frame order, enabling developers to observe the fused scene video in conjunction with the behavior feature table, thereby improving the accuracy of judging the difference between the simulation scene results before and after the autonomous vehicle is given behavior features.

[0121] The A / B testing apparatus for realizing the behavioral characteristics of autonomous vehicles provided in the embodiments of this application will be described below. The A / B testing apparatus for realizing the behavioral characteristics of autonomous vehicles described below can be referred to in correspondence with the A / B testing method for realizing the behavioral characteristics of autonomous vehicles described above.

[0122] See Figure 2 , Figure 2 This is a schematic diagram of an A / B testing device for realizing the behavioral characteristics of autonomous vehicles, as disclosed in an embodiment of this application.

[0123] like Figure 2 As shown, the device may include:

[0124] The road test data acquisition unit 11 is used to acquire first vehicle road test data of an autonomous vehicle with added behavioral features, and second vehicle road test data of an autonomous vehicle without added behavioral features.

[0125] The scene video acquisition unit 12 is used to perform scene simulation on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video, wherein the number of frames of the second vehicle scene video is the same as the number of frames of the first vehicle scene video.

[0126] The scene video fusion unit 13 is used to fuse the second vehicle scene video with the first vehicle scene video as the reference video to obtain a fused scene video, so as to display the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video.

[0127] Optionally, the scene video acquisition unit includes:

[0128] The first vehicle scene video acquisition unit is used to input the first vehicle road test data into the scene simulation system and output the first vehicle scene video.

[0129] The second vehicle scene video acquisition unit is used to input the second vehicle road test data into the scene simulation system and output the second vehicle scene video.

[0130] Optionally, the scene video fusion unit includes:

[0131] The first marking unit is used to mark the main vehicle and obstacles in the first vehicle scene video;

[0132] The second marking unit is used to mark the main vehicle and obstacles in the second vehicle scene video, and the marking color of the second marking is different from the marking color of the first marking.

[0133] The first fusion frame determination unit is used to retain all traffic elements, road elements and obstacles in the first target frame for each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, and to superimpose the traffic elements, road elements and obstacles in the second target frame that are different from those in the first target frame onto the first target frame to obtain the first fusion frame of the first target frame.

[0134] The second fusion frame determination unit is used to, for each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, when the coordinate position of the main vehicle in the second target frame is different from the coordinate position of the main vehicle in the first target frame, superimpose the main vehicle in the second target frame onto the first fusion frame to obtain the second fusion frame of the first target frame.

[0135] The video synthesis unit is used to synthesize the second fusion frames of each first target frame to obtain a fused scene video.

[0136] Optionally, the device may also include:

[0137] The main vehicle information acquisition unit is used to acquire the indicator light switch status information, gear shift information and acceleration / deceleration information of the main vehicle in each first target frame, and the indicator light switch status information, gear shift information and acceleration / deceleration information of the main vehicle in each second target frame.

[0138] The third marking unit is used to mark the main vehicle in the first target frame and the main vehicle in the second target frame in the second vehicle scene video with a third mark if the indicator light status information of the main vehicle in the first target frame is different from that of the main vehicle in the second target frame. The marking color of the third mark is different from that of the marking color of the first mark and the marking color of the second mark.

[0139] Optionally, the device may also include:

[0140] The fourth marking unit is used to mark the main vehicle in the first target frame and the main vehicle in the second target frame in the second vehicle scene video with a fourth mark if the gear shifting information of the main vehicle in the first target frame is different from that of the main vehicle in the second target frame. The marking color of the fourth mark is different from the marking color of the first mark, the marking color of the second mark and the marking color of the third mark.

[0141] Optionally, the device may also include:

[0142] The fifth marking unit is used to mark the main vehicle in the first target frame and the main vehicle in the second target frame in the second vehicle scene video with a fifth mark if the acceleration and deceleration information of the main vehicle in the first target frame is different from that in the second target frame. The marking color of the fifth mark is different from the marking color of the first mark, the marking color of the second mark, the marking color of the third mark and the marking color of the fourth mark.

[0143] Optionally, the first marking unit includes:

[0144] The first vehicle obstacle marking unit is used to mark the main vehicle and the target obstacle in each frame of the first vehicle scene video when the distance between the main vehicle and the target obstacle in the frame is less than a preset distance.

[0145] Optionally, the second marking unit includes:

[0146] The second vehicle obstacle marking unit is used to mark the main vehicle and the target obstacle in each frame of the second vehicle scene video when the distance between the main vehicle and the target obstacle in the frame is less than a preset distance.

[0147] Optionally, the device may also include:

[0148] The first main vehicle information acquisition unit is used to acquire the speed information, acceleration information and steering wheel information of the main vehicle in each frame of the first vehicle scene video.

[0149] The second main vehicle information acquisition unit is used to acquire the speed information, acceleration information and steering wheel information of the main vehicle in each frame of the second vehicle scene video.

[0150] The behavior feature table construction unit is used to construct a behavior feature table based on the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the first vehicle scene video, and the speed information, acceleration information, and steering wheel steering information of the main vehicle in each frame of the second vehicle scene video.

[0151] The A / B testing device for the behavioral characteristics of autonomous vehicles provided in this application embodiment can be applied to devices for A / B testing of the behavioral characteristics of autonomous vehicles, such as computers, servers, and cloud computing. Optionally, Figure 3 The diagram shows the hardware structure of an apparatus for A / B testing of the behavioral characteristics of autonomous vehicles. Figure 3 The hardware structure of the device for A / B testing of the behavioral characteristics of autonomous vehicles may include: at least one processor 1, at least one communication interface 2, at least one memory 3, and at least one communication bus 4.

[0152] In this embodiment of the application, the number of processor 1, communication interface 2, memory 3, and communication bus 4 is at least one, and processor 1, communication interface 2, and memory 3 communicate with each other through communication bus 4;

[0153] Processor 1 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention.

[0154] Memory 3 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk storage device;

[0155] The memory stores a program, which the processor can call. The program is used for:

[0156] Acquire first vehicle road test data of an autonomous vehicle with added behavioral features, and second vehicle road test data of an autonomous vehicle without added behavioral features;

[0157] Scene simulations were performed on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video. The number of frames in the second vehicle scene video was the same as the number of frames in the first vehicle scene video.

[0158] Using the first vehicle scene video as a reference video, the second vehicle scene video is fused to obtain a fused scene video, which displays the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video.

[0159] Optionally, the refined and extended functions of the program can be found in the description above.

[0160] This application embodiment also provides a storage medium that can store a program suitable for execution by a processor, the program being used for:

[0161] Acquire first vehicle road test data of an autonomous vehicle with added behavioral features, and second vehicle road test data of an autonomous vehicle without added behavioral features;

[0162] Scene simulations were performed on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video. The number of frames in the second vehicle scene video was the same as the number of frames in the first vehicle scene video.

[0163] Using the first vehicle scene video as a reference video, the second vehicle scene video is fused to obtain a fused scene video, which displays the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video.

[0164] Optionally, the refined and extended functions of the program can be found in the description above.

[0165] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0166] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0167] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An A / B testing method for behavioral characteristics of autonomous vehicles, characterized in that, include: Acquire first vehicle road test data of an autonomous vehicle with added behavioral features, and second vehicle road test data of an autonomous vehicle without added behavioral features; Scene simulations were performed on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video. The number of frames in the second vehicle scene video was the same as the number of frames in the first vehicle scene video. Using the first vehicle scene video as a reference video, the second vehicle scene video is fused to obtain a fused scene video, which displays the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video. Using the first vehicle scene video as a reference video, the second vehicle scene video is fused to obtain a fused scene video, including: The main vehicle and obstacles in the first vehicle scene video are marked first; The main vehicle and obstacles in the second vehicle scene video are marked with a second mark, and the mark color of the second mark is different from the mark color of the first mark; For each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, all traffic elements, road elements and obstacles in the first target frame are retained, and traffic elements, road elements and obstacles in the second target frame that are different from those in the first target frame are superimposed on the first target frame to obtain the first fused frame of the first target frame. For each first target frame of the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, when the coordinate position of the main vehicle in the second target frame is different from the coordinate position of the main vehicle in the first target frame, the main vehicle in the second target frame is superimposed on the first target frame to obtain the second target frame. The second fusion frames of each first target frame are combined to obtain a fused scene video.

2. The method according to claim 1, characterized in that, The step of performing scene simulations on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video includes: The road test data of the first vehicle is input into the scene simulation system, and the scene video of the first vehicle is output. The second vehicle road test data is input into the scene simulation system, and the second vehicle scene video is output.

3. The method according to claim 2, characterized in that, Also includes: Acquire the indicator light switch status information, gear shift information, and acceleration / deceleration information of the main vehicle in each first target frame, and the indicator light switch status information, gear shift information, and acceleration / deceleration information of the main vehicle in each second target frame; For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the indicator light status information of the main vehicle in the first target frame is different from the indicator light status information of the main vehicle in the second target frame, a third mark is marked on both the main vehicle in the first target frame and the main vehicle in the second target frame. The mark color of the third mark is different from the mark color of the first mark and the mark color of the second mark.

4. The method according to claim 3, characterized in that, Also includes: For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the gear shifting information of the main vehicle in the first target frame is different from the gear shifting information of the main vehicle in the second target frame, a fourth mark is marked on both the main vehicle in the first target frame and the main vehicle in the second target frame. The mark color of the fourth mark is different from the mark color of the first mark, the mark color of the second mark and the mark color of the third mark.

5. The method according to claim 4, characterized in that, Also includes: For each first target frame in the first vehicle scene video and the second target frame in the second vehicle scene video corresponding to the first target frame, if the acceleration / deceleration information of the main vehicle in the first target frame is different from that of the main vehicle in the second target frame, a fifth mark is marked on both the main vehicle in the first target frame and the main vehicle in the second target frame. The mark color of the fifth mark is different from the mark color of the first mark, the mark color of the second mark, the mark color of the third mark, and the mark color of the fourth mark.

6. The method according to claim 1, characterized in that, The first marking is performed on the main vehicle and obstacles in the first vehicle scene video, including: In each frame of the first vehicle scene video, when the distance between the main vehicle and the target obstacle in that frame is less than a preset spacing, the main vehicle and the target obstacle in that frame are marked for the first time.

7. The method according to claim 1, characterized in that, The main vehicle and obstacles in the second vehicle scene video are marked a second time, including: In each frame of the second vehicle scene video, when the distance between the main vehicle and the target obstacle in that frame is less than a preset spacing, the main vehicle and the target obstacle in that frame are marked a second time.

8. The method according to any one of claims 1-7, characterized in that, Also includes: Obtain the speed information, acceleration information, and steering wheel information of the main vehicle in each frame of the first vehicle scene video; Obtain the speed information, acceleration information, and steering wheel information of the main vehicle in each frame of the second vehicle scene video; A behavior feature table is constructed based on the speed, acceleration, and steering information of the main vehicle in each frame of the first vehicle scene video, and the speed, acceleration, and steering information of the main vehicle in each frame of the second vehicle scene video.

9. An A / B testing device for the behavioral characteristics of an autonomous vehicle, characterized in that, The apparatus, applied to the A / B testing method for the behavioral characteristics of autonomous vehicles as described in claim 1, comprises: The road test data acquisition unit is used to acquire first vehicle road test data of an autonomous vehicle with added behavioral features, and second vehicle road test data of an autonomous vehicle without added behavioral features. The scene video acquisition unit is used to perform scene simulation on the first vehicle road test data and the second vehicle road test data respectively to obtain the first vehicle scene video and the second vehicle scene video, wherein the number of frames of the second vehicle scene video is the same as the number of frames of the first vehicle scene video. The scene video fusion unit is used to fuse the second vehicle scene video with the first vehicle scene video as the reference video to obtain a fused scene video, so as to display the behavioral differences between the main vehicle in the first vehicle scene video and the main vehicle in the second vehicle scene video.

10. An A / B testing device for the behavioral characteristics of an autonomous vehicle, characterized in that, Including memory and processor; The memory is used to store programs; The processor is configured to execute the program to implement the steps of the A / B testing method for the behavioral characteristics of an autonomous vehicle as described in any one of claims 1-8.

11. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the A / B testing method for the behavioral characteristics of an autonomous vehicle as described in any one of claims 1-8.

Images